I recently upgraded from a GTX 1650 to an RTX 3070 so I could play newer games at better settings, even though my i5-9400F may bottleneck it a little. While playing Overwatch, the display suddenly went black and the GPU fans ramped up loudly. I could still hear my friends in voice chat, so the rest of the system seemed to be running. After restarting, I installed a GPU power-limiting tool because I was worried about damaging the card. Could this be caused by the graphics card or power supply? My system has a Vero L3 600W PSU using a single PCIe cable with a splitter, an RTX 3070, an i5-9400F, and 16GB of DDR4 memory.
4 Answers
The card itself could be faulty, especially if it was purchased used, but black video combined with fans suddenly running at maximum is not enough to prove that the GPU is dying. Test the system with a known-good, adequately rated PSU and use separate PCIe cables if possible. If the problem continues with a reliable PSU and clean drivers, then the graphics card becomes much more suspicious.
Since you were already using an Nvidia card, a driver mismatch is less likely, but it is still worth completely removing the old driver with DDU and installing a fresh RTX 3070 driver. Also check GPU temperatures and event logs after the next crash. If temperatures are excessive, the card could have a cooling problem, although I would investigate the PSU and cabling before replacing thermal paste.
That symptom often points to the power supply or the PCIe power connection. A sudden load spike can cause the GPU to stop responding while the fans go to full speed, even though Windows and voice chat continue running. A 600W unit may be borderline depending on its quality, and the 3070 should ideally use two separate PCIe power cables rather than one daisy-chained cable. Power limiting can reduce the chance of triggering the problem, but it does not fix an inadequate or failing PSU.
It is currently using one cable with a splitter, so I’m starting to think the PSU is the likely problem.
Before changing anything inside the GPU, monitor temperatures under load and make sure the card is seated properly. Check that the power connectors are fully inserted and inspect them for damage or unusual heat. Avoid repeatedly gaming with the issue until the power delivery has been checked.
I already updated the drivers, and I’m not very comfortable opening the card to replace the thermal paste unless it becomes necessary.